[PD-dev] ugly segfaults with external...

Ben Bogart - FMPM/F1999 bbogart at acs.ryerson.ca
Wed Jul 24 19:38:58 CEST 2002

Hi all,

I seem to be getting some very ugly segfaults from strange interactions 
between my chaos math externals (lorenz, henon, rossler).

Each external works perfectly on thier own, but if I put more than one in 
a patch, or changle the name of one to another in an object box I get 
strange values output (like + or nan) , or segfaults. 

I'm just sending the henon external, the other two work just the same but 
with different equations. 

Here is my Source:

#include "m_pd.h"
#include <math.h>
#include <stdlib.h>

t_class *myclass;

typedef struct thisismystruct
        t_object myobj;
        double a,b,lx0,ly0;
        t_outlet *y_outlet;
}       mystruct;

void calculate(mystruct *x)
        double lx0,ly0,lx1,ly1;
        double a,b;

        a = x->a;
        b = x->b;
        lx0 = x->lx0;
        ly0 = x->ly0;

        lx1 = (ly0 + 1) - (a * pow(lx0,2));
        ly1 = b * lx0;
        x->lx0 = lx1;
        x->ly0 = ly1;

        outlet_float(x->myobj.ob_outlet, (t_float)lx1);
        outlet_float(x->y_outlet, (t_float)ly1);

void reset(mystruct *x, t_floatarg a, t_floatarg b)
        x->a = (double)a;
        x->b = (double)b;
        x->lx0 = 1;
        x->ly0 = 1;

void *henon_new(void)
        mystruct *x = (mystruct *)pd_new(myclass);
        x->a = 1.4;
        x->b = 0.3;
        x->lx0 = 1;
        x->ly0 = 1;

        outlet_new(&x->myobj, &s_float);                /* Default float 
outlet */
        x->y_outlet = outlet_new(&x->myobj, &s_float);
        return (void *)x;

void henon_setup(void)
        myclass = class_new(gensym("henon"),    /* symname is the 
symbolic name */
        (t_newmethod)henon_new,                 /* Constructor Function */
        0,                                              /* Destructor 
Function */
        sizeof(mystruct),                               /* Size of the 
structure */
        CLASS_DEFAULT,                                  /* Graphical 
Representation */
        0);                                             /* 0 Terminates 
Argument List */

        class_addbang(myclass, (t_method)calculate);

Here is my GDB Output:

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 1024 (LWP 5262)]
0x0808a0d7 in outlet_float (x=0x80e04e0, f=-0) at m_obj.c:343
343             pd_float(oc->oc_to, f);
(gdb) where
#0  0x0808a0d7 in outlet_float (x=0x80e04e0, f=-0) at m_obj.c:343
#1  0x400175a0 in calculate () from 
#2  0x0808a007 in outlet_bang (x=0x80e0288) at m_obj.c:311
#3  0x08072473 in bng_bout2 (x=0x80e01b8) at g_7_guis.c:1211
#4  0x08072681 in bng_click (x=0x80e01b8, xpos=184, ypos=115, shift=0, 
    alt=0) at g_7_guis.c:1254
#5  0x080726ce in bng_newclick (z=0x80e01b8, glist=0x80dfc00, xpix=184, 
    shift=0, alt=0, dbl=0, doit=1) at g_7_guis.c:1260
#6  0x0806a54d in gobj_click (x=0x80e01b8, glist=0x80dfc00, xpix=184, 
    shift=0, alt=0, dbl=0, doit=1) at g_editor.c:62
#7  0x0806b5e9 in canvas_doclick (x=0x80dfc00, xpos=184, ypos=115, 
which=1, mod=0, 
    doit=1) at g_editor.c:455
#8  0x0806bd84 in canvas_mousedown (x=0x80dfc00, xpos=184, ypos=115, 
which=1, mod=0)
    at g_editor.c:606
#9  0x08089012 in pd_typedmess (x=0x80dfc00, s=0x80d41d0, argc=0, 
    at m_class.c:669
#10 0x080681c7 in guiconnect_anything (x=0x80dfad8, s=0x80d41d0, ac=4, 
    at g_guiconnect.c:70
#11 0x08089218 in pd_typedmess (x=0x80dfad8, s=0x80d41d0, argc=4, 
    at m_class.c:690
#12 0x0808c984 in binbuf_eval (x=0x80de808, target=0x80dfad8, argc=0, 
    at m_binbuf.c:570
#13 0x08091ca4 in socketreceiver_read (x=0x80de828, fd=8) at s_inter.c:307
#14 0x0809164a in sys_domicrosleep (microsec=0, pollem=1) at s_inter.c:144
#15 0x08092815 in sys_pollgui () at s_inter.c:732
#16 0x0808fee6 in m_scheduler (nodacs=0) at m_sched.c:441
#17 0x080904f9 in sys_main (argc=1, argv=0xbffff7e4) at s_main.c:258
#18 0x08094356 in main (argc=1, argv=0xbffff7e4) at s_entry.c:9
#19 0x4007a9ed in __libc_start_main () from /lib/libc.so.6

Thanks for your help, I would appriciate any help/pointers anyone would 


B. Bogart

More information about the Pd-dev mailing list